Abstract The relationship between marijuana, tobacco, and e-cigarettes is both rapidly changing and poorly understood, particularly in the light of recent federal and state-level regulatory changes governing the availability of e- cigarettes and marijuana products (respectively). In order to understand this changing landscape we need new, ?exible, and responsive research methods capable of rapidly providing insights into product initiation patterns, use patterns, and cessation strategies. Social media ? here de?ned as including internet discussion forums ? provides a ready-made source of abundant, naturalistic, longitudinal, publicly accessible, ?rst-person narratives with which to understand health behaviours and attitudes. We propose to use a combination of qualitative methods and automated natural language processing techniques to investigate online discussion forums devoted to tobacco, marijuana, and e-cigarettes in order to understand user trajectories through the three product categories.